arXiv subjects

Meng Zheng

Publications and source records attributed to Meng Zheng.

At least 55 records · Page 3Linked to original sources

arXiv2023

PREF: Predictability Regularized Neural Motion Fields

Knowing the 3D motions in a dynamic scene is essential to many vision applications. Recent progress is mainly focused on estimating the activity of some specific elements like humans. In this paper, we leverage a neural motion field for estimating the motion of all points in a multiview setting. Modeling the motion from a dynamic scene with multiview data is challenging due to the ambiguities in points of similar color and points with time-varying color. We propose to regularize the estimated motion to be predictable. If the motion from previous frames is known, then the motion in the near future should be predictable. Therefore, we introduce a predictability regularization by first conditioning the estimated motion on latent embeddings, then by adopting a predictor network to enforce predictability on the embeddings. The proposed framework PREF (Predictability REgularized Fields) achieves on par or better results than state-of-the-art neural motion field-based dynamic scene representation methods, while requiring no prior knowledge of the scene.

cs.CV↗

arXiv2022

Progressive Multi-view Human Mesh Recovery with Self-Supervision

To date, little attention has been given to multi-view 3D human mesh estimation, despite real-life applicability (e.g., motion capture, sport analysis) and robustness to single-view ambiguities. Existing solutions typically suffer from poor generalization performance to new settings, largely due to the limited diversity of image-mesh pairs in multi-view training data. To address this shortcoming, people have explored the use of synthetic images. But besides the usual impact of visual gap between rendered and target data, synthetic-data-driven multi-view estimators also suffer from overfitting to the camera viewpoint distribution sampled during training which usually differs from real-world distributions. Tackling both challenges, we propose a novel simulation-based training pipeline for multi-view human mesh recovery, which (a) relies on intermediate 2D representations which are more robust to synthetic-to-real domain gap; (b) leverages learnable calibration and triangulation to adapt to more diversified camera setups; and (c) progressively aggregates multi-view information in a canonical 3D space to remove ambiguities in 2D representations. Through extensive benchmarking, we demonstrate the superiority of the proposed solution especially for unseen in-the-wild scenarios.

cs.CV↗

arXiv2022

Federated Learning with Privacy-Preserving Ensemble Attention Distillation

Federated Learning (FL) is a machine learning paradigm where many local nodes collaboratively train a central model while keeping the training data decentralized. This is particularly relevant for clinical applications since patient data are usually not allowed to be transferred out of medical facilities, leading to the need for FL. Existing FL methods typically share model parameters or employ co-distillation to address the issue of unbalanced data distribution. However, they also require numerous rounds of synchronized communication and, more importantly, suffer from a privacy leakage risk. We propose a privacy-preserving FL framework leveraging unlabeled public data for one-way offline knowledge distillation in this work. The central model is learned from local knowledge via ensemble attention distillation. Our technique uses decentralized and heterogeneous local data like existing FL approaches, but more importantly, it significantly reduces the risk of privacy leakage. We demonstrate that our method achieves very competitive performance with more robust privacy preservation based on extensive experiments on image classification, segmentation, and reconstruction tasks.

cs.LG↗

arXiv2022

Self-supervised Human Mesh Recovery with Cross-Representation Alignment

Fully supervised human mesh recovery methods are data-hungry and have poor generalizability due to the limited availability and diversity of 3D-annotated benchmark datasets. Recent progress in self-supervised human mesh recovery has been made using synthetic-data-driven training paradigms where the model is trained from synthetic paired 2D representation (e.g., 2D keypoints and segmentation masks) and 3D mesh. However, on synthetic dense correspondence maps (i.e., IUV) few have been explored since the domain gap between synthetic training data and real testing data is hard to address for 2D dense representation. To alleviate this domain gap on IUV, we propose cross-representation alignment utilizing the complementary information from the robust but sparse representation (2D keypoints). Specifically, the alignment errors between initial mesh estimation and both 2D representations are forwarded into regressor and dynamically corrected in the following mesh regression. This adaptive cross-representation alignment explicitly learns from the deviations and captures complementary information: robustness from sparse representation and richness from dense representation. We conduct extensive experiments on multiple standard benchmark datasets and demonstrate competitive results, helping take a step towards reducing the annotation effort needed to produce state-of-the-art models in human mesh estimation.

cs.CV↗

arXiv2022

PseudoClick: Interactive Image Segmentation with Click Imitation

The goal of click-based interactive image segmentation is to obtain precise object segmentation masks with limited user interaction, i.e., by a minimal number of user clicks. Existing methods require users to provide all the clicks: by first inspecting the segmentation mask and then providing points on mislabeled regions, iteratively. We ask the question: can our model directly predict where to click, so as to further reduce the user interaction cost? To this end, we propose {\PseudoClick}, a generic framework that enables existing segmentation networks to propose candidate next clicks. These automatically generated clicks, termed pseudo clicks in this work, serve as an imitation of human clicks to refine the segmentation mask.

cs.CV↗

arXiv2022

Visual Similarity Attention

While there has been substantial progress in learning suitable distance metrics, these techniques in general lack transparency and decision reasoning, i.e., explaining why the input set of images is similar or dissimilar. In this work, we solve this key problem by proposing the first method to generate generic visual similarity explanations with gradient-based attention. We demonstrate that our technique is agnostic to the specific similarity model type, e.g., we show applicability to Siamese, triplet, and quadruplet models. Furthermore, we make our proposed similarity attention a principled part of the learning process, resulting in a new paradigm for learning similarity functions. We demonstrate that our learning mechanism results in more generalizable, as well as explainable, similarity models. Finally, we demonstrate the generality of our framework by means of experiments on a variety of tasks, including image retrieval, person re-identification, and low-shot semantic segmentation.

cs.CV↗

arXiv2021

Spatio-Temporal Representation Factorization for Video-based Person Re-Identification

Despite much recent progress in video-based person re-identification (re-ID), the current state-of-the-art still suffers from common real-world challenges such as appearance similarity among various people, occlusions, and frame misalignment. To alleviate these problems, we propose Spatio-Temporal Representation Factorization (STRF), a flexible new computational unit that can be used in conjunction with most existing 3D convolutional neural network architectures for re-ID. The key innovations of STRF over prior work include explicit pathways for learning discriminative temporal and spatial features, with each component further factorized to capture complementary person-specific appearance and motion information. Specifically, temporal factorization comprises two branches, one each for static features (e.g., the color of clothes) that do not change much over time, and dynamic features (e.g., walking patterns) that change over time. Further, spatial factorization also comprises two branches to learn both global (coarse segments) as well as local (finer segments) appearance features, with the local features particularly useful in cases of occlusion or spatial misalignment. These two factorization operations taken together result in a modular architecture for our parameter-wise light STRF unit that can be plugged in between any two 3D convolutional layers, resulting in an end-to-end learning framework. We empirically show that STRF improves performance of various existing baseline architectures while demonstrating new state-of-the-art results using standard person re-ID evaluation protocols on three benchmarks.

cs.CV↗

arXiv2021

Everybody Is Unique: Towards Unbiased Human Mesh Recovery

We consider the problem of obese human mesh recovery, i.e., fitting a parametric human mesh to images of obese people. Despite obese person mesh fitting being an important problem with numerous applications (e.g., healthcare), much recent progress in mesh recovery has been restricted to images of non-obese people. In this work, we identify this crucial gap in the current literature by presenting and discussing limitations of existing algorithms. Next, we present a simple baseline to address this problem that is scalable and can be easily used in conjunction with existing algorithms to improve their performance. Finally, we present a generalized human mesh optimization algorithm that substantially improves the performance of existing methods on both obese person images as well as community-standard benchmark datasets. A key innovation of this technique is that it does not rely on supervision from expensive-to-create mesh parameters. Instead, starting from widely and cheaply available 2D keypoints annotations, our method automatically generates mesh parameters that can in turn be used to re-train and fine-tune any existing mesh estimation algorithm. This way, we show our method acts as a drop-in to improve the performance of a wide variety of contemporary mesh estimation methods. We conduct extensive experiments on multiple datasets comprising both standard and obese person images and demonstrate the efficacy of our proposed techniques.

cs.CV↗

arXiv2021

A Peek Into the Reasoning of Neural Networks: Interpreting with Structural Visual Concepts

Despite substantial progress in applying neural networks (NN) to a wide variety of areas, they still largely suffer from a lack of transparency and interpretability. While recent developments in explainable artificial intelligence attempt to bridge this gap (e.g., by visualizing the correlation between input pixels and final outputs), these approaches are limited to explaining low-level relationships, and crucially, do not provide insights on error correction. In this work, we propose a framework (VRX) to interpret classification NNs with intuitive structural visual concepts. Given a trained classification model, the proposed VRX extracts relevant class-specific visual concepts and organizes them using structural concept graphs (SCG) based on pairwise concept relationships. By means of knowledge distillation, we show VRX can take a step towards mimicking the reasoning process of NNs and provide logical, concept-level explanations for final model decisions. With extensive experiments, we empirically show VRX can meaningfully answer "why" and "why not" questions about the prediction, providing easy-to-understand insights about the reasoning process. We also show that these insights can potentially provide guidance on improving NN's performance.

cs.CV↗

arXiv2020

Towards Visually Explaining Similarity Models

We consider the problem of visually explaining similarity models, i.e., explaining why a model predicts two images to be similar in addition to producing a scalar score. While much recent work in visual model interpretability has focused on gradient-based attention, these methods rely on a classification module to generate visual explanations. Consequently, they cannot readily explain other kinds of models that do not use or need classification-like loss functions (e.g., similarity models trained with a metric learning loss). In this work, we bridge this crucial gap, presenting a method to generate gradient-based visual attention for image similarity predictors. By relying solely on the learned feature embedding, we show that our approach can be applied to any kind of CNN-based similarity architecture, an important step towards generic visual explainability. We show that our resulting attention maps serve more than just interpretability; they can be infused into the model learning process itself with new trainable constraints. We show that the resulting similarity models perform, and can be visually explained, better than the corresponding baseline models trained without these constraints. We demonstrate our approach using extensive experiments on three different kinds of tasks: generic image retrieval, person re-identification, and low-shot semantic segmentation.

cs.CV↗

arXiv2020

Compressive Subspace Learning with Antenna Cross-correlations for Wideband Spectrum Sensing

Compressive subspace learning (CSL) with the exploitation of space diversity has found a potential performance improvement for wideband spectrum sensing (WBSS). However, previous works mainly focus on either exploiting antenna auto-correlations or adopting a multiple-input multiple-output (MIMO) channel without considering the spatial correlations, which will degrade their performances. In this paper, we consider a spatially correlated MIMO channel and propose two CSL algorithms (i.e., mCSLSACC and vCSLACC) which exploit antenna cross-correlations, where the mCSLSACC utilizes an antenna averaging temporal decomposition, and the vCSLACC uses a spatial-temporal joint decomposition. For both algorithms, the conditions of statistical covariance matrices (SCMs) without noise corruption are derived. Through establishing the singular value relation of SCMs in statistical sense between the proposed and traditional CSL algorithms, we show the superiority of the proposed CSL algorithms. By further depicting the receiving correlation matrix of MIMO channel with the exponential correlation model, we give important closed-form expressions for the proposed CSL algorithms in terms of the amplification of singular values over traditional CSL algorithms. Such expressions provide a possibility to determine optimal algorithm parameters for high system performances in an analytical way. Simulations validate the correctness of this work and its performance improvement over existing works in terms of WBSS performance.

cs.IT↗

arXiv2020

Towards Visually Explaining Variational Autoencoders

Recent advances in Convolutional Neural Network (CNN) model interpretability have led to impressive progress in visualizing and understanding model predictions. In particular, gradient-based visual attention methods have driven much recent effort in using visual attention maps as a means for visual explanations. A key problem, however, is these methods are designed for classification and categorization tasks, and their extension to explaining generative models, e.g. variational autoencoders (VAE) is not trivial. In this work, we take a step towards bridging this crucial gap, proposing the first technique to visually explain VAEs by means of gradient-based attention. We present methods to generate visual attention from the learned latent space, and also demonstrate such attention explanations serve more than just explaining VAE predictions. We show how these attention maps can be used to localize anomalies in images, demonstrating state-of-the-art performance on the MVTec-AD dataset. We also show how they can be infused into model training, helping bootstrap the VAE into learning improved latent space disentanglement, demonstrated on the Dsprites dataset.

cs.CV↗

arXiv2019

Re-Identification with Consistent Attentive Siamese Networks

We propose a new deep architecture for person re-identification (re-id). While re-id has seen much recent progress, spatial localization and view-invariant representation learning for robust cross-view matching remain key, unsolved problems. We address these questions by means of a new attention-driven Siamese learning architecture, called the Consistent Attentive Siamese Network. Our key innovations compared to existing, competing methods include (a) a flexible framework design that produces attention with only identity labels as supervision, (b) explicit mechanisms to enforce attention consistency among images of the same person, and (c) a new Siamese framework that integrates attention and attention consistency, producing principled supervisory signals as well as the first mechanism that can explain the reasoning behind the Siamese framework's predictions. We conduct extensive evaluations on the CUHK03-NP, DukeMTMC-ReID, and Market-1501 datasets and report competitive performance.

cs.CV↗

arXiv2018

Measuring the Temporal Behavior of Real-World Person Re-Identification

Designing real-world person re-identification (re-id) systems requires attention to operational aspects not typically considered in academic research. Typically, the probe image or image sequence is matched to a gallery set with a fixed candidate list. On the other hand, in real-world applications of re-id, we would search for a person of interest in a gallery set that is continuously populated by new candidates over time. A key question of interest for the operator of such a system is: how long is a correct match to a probe likely to remain in a rank-k shortlist of candidates? In this paper, we propose to distill this information into what we call a Rank Persistence Curve (RPC), which unlike a conventional cumulative match characteristic (CMC) curve helps directly compare the temporal performance of different re-id algorithms. To carefully illustrate the concept, we collected a new multi-shot person re-id dataset called RPIfield. The RPIfield dataset is constructed using a network of 12 cameras with 112 explicitly time-stamped actor paths among about 4000 distractors. We then evaluate the temporal performance of different re-id algorithms using the proposed RPCs using single and pairwise camera videos from RPIfield, and discuss considerations for future research.

cs.CV↗

arXiv2017

Deep Regression Bayesian Network and Its Applications

Deep directed generative models have attracted much attention recently due to their generative modeling nature and powerful data representation ability. In this paper, we review different structures of deep directed generative models and the learning and inference algorithms associated with the structures. We focus on a specific structure that consists of layers of Bayesian Networks due to the property of capturing inherent and rich dependencies among latent variables. The major difficulty of learning and inference with deep directed models with many latent variables is the intractable inference due to the dependencies among the latent variables and the exponential number of latent variable configurations. Current solutions use variational methods often through an auxiliary network to approximate the posterior probability inference. In contrast, inference can also be performed directly without using any auxiliary network to maximally preserve the dependencies among the latent variables. Specifically, by exploiting the sparse representation with the latent space, max-max instead of max-sum operation can be used to overcome the exponential number of latent configurations. Furthermore, the max-max operation and augmented coordinate ascent are applied to both supervised and unsupervised learning as well as to various inference. Quantitative evaluations on benchmark datasets of different models are given for both data representation and feature learning tasks.

cs.LG↗

arXiv2017

End-to-end Throughput Maximization for Underlay Multi-hop Cognitive Radio Networks with RF Energy Harvesting

This paper studies a green paradigm for the underlay coexistence of primary users (PUs) and secondary users (SUs) in energy harvesting cognitive radio networks (EH-CRNs), wherein battery-free SUs capture both the spectrum and the energy of PUs to enhance spectrum efficiency and green energy utilization. To lower the transmit powers of SUs, we employ multi-hop transmission with time division multiple access, by which SUs first harvest energy from the RF signals of PUs and then transmit data in the allocated time concurrently with PUs, all in the licensed spectrum. In this way, the available transmit energy of each SU mainly depends on the harvested energy before the turn to transmit, namely energy causality. Meanwhile, the transmit powers of SUs must be strictly controlled to protect PUs from harmful interference. Thus, subject to the energy causality constraint and the interference power constraint, we study the end-to-end throughput maximization problem for optimal time and power allocation. To solve this nonconvex problem, we first equivalently transform it into a convex optimization problem and then propose the joint optimal time and power allocation (JOTPA) algorithm that iteratively solves a series of feasibility problems until convergence. Extensive simulations evaluate the performance of EH-CRNs with JOTPA in three typical deployment scenarios and validate the superiority of JOTPA by making comparisons with two other resource allocation algorithms.

cs.IT↗

arXiv2015

A Connectivity-Aware Approximation Algorithm for Relay Node Placement in Wireless Sensor Networks

In two-tiered Wireless Sensor Networks (WSNs) relay node placement is one of the key factors impacting the network energy consumption and the system overhead. In this paper, a novel connectivity-aware approximation algorithm for relay node placement in WSNs is proposed to offer a major step forward in saving system overhead. Specifically, a unique Local Search Approximation Algorithm (LSAA) is introduced to solve the Relay Node Single Cover (RNSC) problem. In this proposed LSAA approach, the sensor nodes are allocated into groups and then a local Set Cover (SC) for each group is achieved by a local search algorithm. The union set of all local SCs constitutes a SC of the RNSC problem. The approximation ratio and the time complexity of the LSAA are analyzed by rigorous proof. Additionally, the LSAA approach has been extended to solve the relay node double cover problem. Then, a Relay Location Selection Algorithm (RLSA) is proposed to utilize the resulting SC from LSAA in combining RLSA with the minimum spanning tree heuristic to build the high-tier connectivity. As the RLSA searches for a nearest location to the sink node for each relay node, the high-tier network built by RLSA becomes denser than that by existing works. As a result, the number of added relay nodes for building the connectivity of the high-tier WSN can be significantly saved. Simulation results clearly demonstrate that the proposed LSAA outperforms the approaches reported in literature and the RLSA-based algorithm can noticeably save relay nodes newly deployed for the high-tier connectivity.

cs.NI↗

arXiv2015

Set Covering-based Approximation Algorithm for Delay Constrained Relay Node Placement in Wireless Sensor Networks

The Delay Constrained Relay Node Placement (DCRNP) problem in Wireless Sensor Networks (WSNs) aims to deploy minimum relay nodes such that for each sensor node there is a path connecting this sensor node to the sink without violating delay constraint. As WSNs are gradually employed in time-critical applications, the importance of the DCRNP problem becomes noticeable. For the NP-hard nature of DCRNP problem, an approximation algorithm-Set-Covering-based Relay Node Placement (SCA) is proposed to solve the DCRNP problem in this paper. The proposed SCA algorithm deploys relay nodes iteratively from sink to the given sensor nodes in hops, i.e., in the $k$th iteration SCA deploys relay nodes at the locations that are $k$ hops apart from the sink. Specifically, in each iteration, SCA first finds the candidate deployment locations located within 1 hop to the relay nodes and sensor nodes, which have already been connected to the sink. Then, a subset of these candidate deployment locations, which can guarantee the existence of paths connecting unconnected sensor nodes to the sink within delay constraint, is selected to deploy relay nodes based on the set covering method. As the iteration of SCA algorithm, the sensor nodes are gradually connected to the sink with satisfying delay constraint. The elaborated analysis of the approximation ratio of SCA algorithm is given out, and we also prove that the SCA is a polynomial time algorithm through rigorous time complexity analysis. To evaluate the performance of the proposed SCA algorithm, extensive simulations are implemented, and the simulation results show that the SCA algorithm can significantly save the deployed relay nodes comparing to the existing algorithms, i.e., at most 31.48% deployed relay nodes can be saved due to SCA algorithm.

cs.NI↗